To investigate variations in prostate-specific antigen (PSA) levels among men with an initial normal PSA level in the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial study.

Data were extracted from the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial study data set on all men in the interventional arm, with 2 tests performed in a period of <2 years and with an initial result of the first test <4 ng/mL. The range of variation between first and second tests was computed. Risks of cancer stratified on Gleason score were computed using logistic regression.

A total of 31,286 men had 2 PSA tests within 2 years and with an initial value <4 ng/mL. From the first to the second test, the median variation of PSA levels was 3.4% (interquartile range, −15% to +26%). The variation in PSA value was not associated with the delay between the first and the second test (P = .36), age (P = .16), body mass index (P = .41), and race (P = .12). A total of 2,781 prostate cancers were diagnosed during follow-up. Adjusting for age and initial PSA level, the risk of prostate cancer increased linearly with increasing PSA level at the second test, with an odds ratio of 1.079 (95% confidence interval, 1.058-1.101) for each percent increase in PSA level. However, the variation in PSA was not associated with a higher Gleason score (P = .95 for level variations in cancer of Gleason score <7 vs ≥7).

Although an increase in PSA level over time is associated with increased risk of prostate cancer, this association is not related to more aggressive tumors.